From clang-tidy bugprone-incorrect-roundings check:
> casting (double + 0.5) to integer leads to incorrect rounding; consider using lround (#include <cmath>) instead